votianng/backend/src/main/java/de/stadtbote/votian/LegacyCronFiles.java

package de.stadtbote.votian;

import org.springframework.stereotype.Component;
import org.springframework.core.env.Environment;
import org.apache.commons.compress.compressors.bzip2.BZip2CompressorOutputStream;
import org.apache.commons.compress.compressors.bzip2.BZip2CompressorInputStream;
import java.io.*;
import java.nio.file.*;
import java.nio.file.attribute.BasicFileAttributes;
import java.nio.channels.FileChannel;
import java.time.*;
import java.time.format.DateTimeFormatter;
import java.util.*;
import static java.nio.file.LinkOption.NOFOLLOW_LINKS;
import static de.stadtbote.votian.AdminSupport.bad;

/** Native do_bzip2.php. Only explicitly mounted legacy paths are eligible, never NG document stores. */
@Component
class LegacyCronFiles {
    final Environment env;
    LegacyCronFiles(Environment env){this.env=env;}
    static List<Path> paths(String value) throws IOException {
        var result=new ArrayList<Path>();
        for(String raw:value.split(",")){if(raw.isBlank())continue;Path p=Path.of(raw.strip()).toAbsolutePath().normalize();
            if(p.getNameCount()<3||!Files.isDirectory(p,NOFOLLOW_LINKS)||!p.equals(p.toRealPath()))throw bad("Cron-Dateipfad muss ein eigenes vorhandenes Verzeichnis ohne symbolische Links sein.");
            result.add(p);
        }return result;
    }
    static boolean same(BasicFileAttributes a,BasicFileAttributes b){return a.size()==b.size()&&a.lastModifiedTime().equals(b.lastModifiedTime())&&Objects.equals(a.fileKey(),b.fileKey());}
    static String digest(InputStream in) throws IOException {try(in){var md=java.security.MessageDigest.getInstance("SHA-256");byte[] b=new byte[65536];int n;while((n=in.read(b))!=-1)md.update(b,0,n);return HexFormat.of().formatHex(md.digest());}catch(java.security.NoSuchAlgorithmException e){throw new IllegalStateException(e);}}
    static void compress(Path source) throws IOException {
        var before=Files.readAttributes(source,BasicFileAttributes.class,NOFOLLOW_LINKS);if(!before.isRegularFile())return;
        Path target=source.resolveSibling(source.getFileName()+".bz2");
        if(Files.exists(target,NOFOLLOW_LINKS)){
            if(!Files.isRegularFile(target,NOFOLLOW_LINKS))throw bad("Cron-Archivziel ist keine reguläre Datei.");
            String original=digest(Files.newInputStream(source,NOFOLLOW_LINKS));
            String archived=digest(new BZip2CompressorInputStream(Files.newInputStream(target,NOFOLLOW_LINKS)));
            if(!original.equals(archived))throw bad("Vorhandenes Cron-Archiv passt nicht zur Quelldatei; kein Überschreiben.");
        }else{
            Path temp=Files.createTempFile(source.getParent(),".ng-cron-",".part");
            try{
                try(var input=Files.newInputStream(source,NOFOLLOW_LINKS);var output=new BZip2CompressorOutputStream(Files.newOutputStream(temp),9)){input.transferTo(output);}
                try(var channel=FileChannel.open(temp,StandardOpenOption.WRITE)){channel.force(true);}
                if(!same(before,Files.readAttributes(source,BasicFileAttributes.class,NOFOLLOW_LINKS)))throw bad("Quelldatei wurde während der Kompression geändert.");
                // Atomic publication without overwrite; the temporary file is on the same filesystem.
                Files.createLink(target,temp);
            }finally{Files.deleteIfExists(temp);}
        }
        if(!same(before,Files.readAttributes(source,BasicFileAttributes.class,NOFOLLOW_LINKS)))throw bad("Quelldatei wurde während der Kompression geändert.");
        Files.delete(source);
    }
    LegacyCronRunner.Result run(LocalDate today){
        try{
            var bin=paths(env.getProperty("app.legacy-cron.files.binlog-directories",""));
            var logs=paths(env.getProperty("app.legacy-cron.files.log-directories",""));
            var purge=paths(env.getProperty("app.legacy-cron.files.temporary-directories",""));
            if(bin.isEmpty()&&logs.isEmpty()&&purge.isEmpty())throw bad("Keine Altdatei-Verzeichnisse für die Cron-Wartung konfiguriert.");
            var all=new ArrayList<Path>();all.addAll(bin);all.addAll(logs);all.addAll(purge);
            for(int i=0;i<all.size();i++)for(int j=i+1;j<all.size();j++)if(all.get(i).startsWith(all.get(j))||all.get(j).startsWith(all.get(i)))throw bad("Cron-Dateipfade dürfen sich nicht überlappen.");
            int count=0;
            for(Path directory:bin){List<Path> candidates;try(var stream=Files.list(directory)){candidates=stream.filter(p->Files.isRegularFile(p,NOFOLLOW_LINKS)&&p.getFileName().toString().matches("mps2-bin\\.0[0-9]+" )).toList();}
                long newest=Long.MIN_VALUE;for(var p:candidates)newest=Math.max(newest,Files.getLastModifiedTime(p,NOFOLLOW_LINKS).toMillis());
                for(var p:candidates)if(Files.getLastModifiedTime(p,NOFOLLOW_LINKS).toMillis()<newest){compress(p);count++;}
            }
            String month="_"+today.minusMonths(1).format(DateTimeFormatter.ofPattern("yyyyMM"));
            for(Path directory:logs){List<Path> candidates;try(var stream=Files.list(directory)){candidates=stream.filter(p->Files.isRegularFile(p,NOFOLLOW_LINKS)&&p.getFileName().toString().contains(month)&&p.getFileName().toString().endsWith(".log")).toList();}for(var p:candidates){compress(p);count++;}}
            // PHP deletes all regular files recursively in these three temporary trees; links stay untouched.
            for(Path directory:purge)try(var stream=Files.walk(directory)){for(var p:stream.filter(p->Files.isRegularFile(p,NOFOLLOW_LINKS)).toList()){Files.delete(p);count++;}}
            return LegacyCronRunner.Result.ok(count);
        }catch(IOException e){return new LegacyCronRunner.Result("UNCERTAIN","Dateiwartung unterbrochen; Quelldateien und Archive vor Freigabe prüfen.",0);}
    }
}